Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
NBER WORKING PAPER SERIES FEDERAL RESERVE BEHAVIOR SINCE 1980: A FINANCIAL MARKETS PERSPECTIVE William C. Melton and V. Vance Roley Working Paper No. 2608 NATIONAL BUREAU OF ECONOMIC RESEARCH 1050 Massachusetts Avenue Cambridge, MA 02138 June 1988 We are grateful to Thomas Mayer and Gordon Sellon for helpful conunents, to Steven R. Thorley for research assistance, and to the National Science This paper was Foundation (Grant No. SES-8408603) for research support. of American MoncttL The Political Economy for Thomas (ed.), Mayer prepared This research is part of NBER's research program in Financial Policy. Narketa and Monetary Economics. Any opinions expressed are those of t.o authors and not those of the National Bureau of Economic Research. NBER Working Paper #2608 June 1988 Federal Reaerve Behavior Since 1980: A Financial Marketa Perapective ABSTRACT The financial market's understanding of Federal Reserve behavior is used to examine recent changes in monetary policy. Changes in the level of interest rates in response to specific types of economic information are primarily considered- Differences in the volatility of interest rates across periods provide additionalevidence on changes in monetary policy regimes. The results indicate that monetary desire to policy changed several times since 1980 with respect to either the Federal Reserve's targets, its achieve its targets, or its operating procedures. The different regimes correspond to Federal Reserve statements about changes in policy. In this context, then, the evidence- suggests that policy was credible. William C. Meltnn IDS Financial Services, IDS Tower 10 Minneapolis, MN 55440 Inc. V. Vance Roley Department of Finance 0.1-10 University of Washington Seattle, WA 98195 FEDERAL RESERVE BEHAVIOR SINCE 1980: A FINANCIAL MARKETS PERSPECTIVE* William C. Melton and V. Vance Roley Knowledge of actual Federal Reserve behavior is important in studies of monetary policy and financial markets for at least two reasons. First, the monetary policy process may be marred if they interpretation of variables chosen to represent the do not correspond to variables actually used by the Federal Reserve to implement monetary policy or to gauge its performance. Accordingly, the Federal Reserve's choices regarding targets, intermediate targets, and instruments may play research design.1 a key role in Unfortunately, information regarding these choices is not always easy to obtain. Relevant Federal Reserve policy statements, such as the FOMC policy directive, are released with a substantial delay and often are ambiguous.2 Thus, additional knowledge regarding actual Federal Reserve behavior may have a methodological payoff.3 Second, the recurrent issue of policy credibility requires an assessment of the extent to which Federal Reserve statements find a reflection in the beliefs and behavior credibility potentially has a variety of economic agents. While of interpretations, all would seem to require that a shxft in stated policy objectives and instruments be associated with at least some change in market behavior. Thus, additional knowledge regarding Federal Reserve behavior, and the financial market's reaction to it, may help illuminate the credibility issue. This paper examines actual Federal Reserve behavior from a financial markets perspective. Movements in interest rates are used as the metric in this exercise. The underlying presumption is that financial market participants fully understand Federal Reserve behavior. This position is sensible for two reasons. First, many financial market participants are former Federal Reserve officials and economists. Second, given the key role the short-run, market participantshave of the Federal Reserve in influencing interest rates in at least a strong incentive to study Federal Reserve behavior. Following this introductory section, a framework for examining monetary policy is presented in the first section. Within this framework, several different types of monetary policy regimes are distinguished. In the second section, the implications of the different monetary policy regimes on the hehavior of interest rates are discussed. The third section presents empirical resolD u sotual Federal Reserve behavior. The period beginning in 1980 and ending in early 1987 is considered, This period is interesting in that Federal Reserve statements indicate that monetary policy changed several times. Thus specific instances of Federal Reserve behavior can be analyzed in terms of the credibility of pohcymakers. The main conclusions are summarized in the final section. I. The Monetary jic Frawrk Several aspects of monetary policy are reviewed in a stylized framework in this section The long-run objectives of monetary oolicy are first considered. These long-run objectives involve both the targets and intermediate targets of policy over a gwen year. Next, several features pertaining to shortrun rr.onetary policy are discussed. it is argued that to interpret Federai Reserve behsvior correctly. the short-run impemertation of monetary policy must be examined. Different types of' short-run policies are distinguished both by the type of operating procedure implemented by the Federal Resent and by the desire to offset deviationsfrom the targets A. Monetary Policy Qçjives The uncertainties facing the Federal Reserve make monetary policy a particularly challenging task, even apart from outside political pressures. The first choice facing .poticymskers is to decide whetF.sr policy should be based directly on inflation, and foreign exchange rates, or on a set of ultimate targets, such an intermediate target. as output, employment. Potential intermediate targets include monetary and credit aggregates, as well as other indicators from financial markets, including interest rates. In the mid to late 1970s, the intermediate target procedure was progressively made more of a formal part of Federal Reserve policy. Coinciding with this adoption, the Federal Reserve initiated a series of annual targets for monetary and credit aggregates. This procedure was formalized further under the Humphrey-Hawkins Act in 1978, which imposed a single non-overlapping calendar year policy period on the Federal Reserve. By the early 1980s, the intermediate target procedure was firmly entrenched .Nevertheless, 2 some ambiguities were apparent. One area of ambiguity involved the appropriate weights to be placed on the various intermediate targets, particularly MI and M2, in formulating policy. A second area involved uncertainties about the proper definition of the narrowly defined money stock, MI. Despite erratic MI problems in determining an appropriate definition for Ml, and the related problem of velocity, Ml was the main focus of monetary policy. One reason for the use of Ml was its timeliness. data on broader Preliminary Mi data are available weekly with a lag of about 1-1/2 weeks, while monetary aggregates are available only monthly. Many of the components are, however, available weekly. A second was of the broader aggregates that Ml was the traditionally preferred monetary aggregate because of its intended link with transactions balances. Another reason was that the reserve requirements applied to non-Mi components of the broader aggregates are either zero or very low, implying a potentially loose relationship between reserves and the broader aggregates. Because of the continuing erratic behavior of Ml velocity, the Federal Reserve virtually abandoned its intermediate target procedure by the mid-1980s. In the record of policy actions at the November 2, 1987 FOMC meeting, for example, the FOMC's domestic policy directive listed the behavior of the monetary aggregates last among four items that could cause the Federal Reserve to charge current pressure on reserve positions. The first three items were the strength of the business expansion, inflationary pressures, and developments in foreign exchange markets. Thus. policymakers were looking directly at measures of economic performance. It also is noteworthy that the items in this list occasionally change order. Developments in foreign exchange markets, for example, had been listed first a few months earlier, The above is clearly a casual interpretation of monetary policy objectives in the 1980s. It does, nevertheless, suggest several hypotheses about Federal Reserve behavior. These hypotheses are tested in the third section. However, to interpret Federal Reserve behavior more precisely, the short-run implementation of policy must be considered. B. Short-Run Monetary j]j The Federal Reserve's short-run monetary policy can be described in terms of two factors. The first concerns the rate at which the Federal Reserve attempts to offset any deviation from its target or 3 intermediate target. The second is the type of operating procedure adopted. This latter area involves the choice of an instrument, or operating target, to conduct monetary policy. The view of monetary policy advanced here is that the Federal Reserve does not initiate short-. run shocks to the monetary aggregates; it merely reacts to them. This view seems particularly appropriate prior to Fehruary 1984, when contemporaneous reserve requirements (CRR) were adopted. Under the lagged reserve requirement (LRR) system in effect from 1968 to that time, there was no direct link between hank reserves and MI in a given week, As a consequence, the money stock was essentially demand determined, and monetary shocks reflected shifts in the public's demand for money.4 Given that observed monetary shocks represent new information to boththe public and the Fcderal Reserve, the relevant issue for policymakers is to determin the desired adjustment toward the monetary target. Casual evidence suggests that this adjustment speed may have changed at least twice since the late 1970s. In particular, the Federal Reserve committed itself more ciosely to Ml targets in October 1979, implying that any deviation from the monetary targets would be offset more quickly than before. In October 1982, the Federal Reserve de-emphssized its monetary targets, implying slower adjustment speeds. At the same time that the Federal Reserve changes adjustment speeds, it may also change its operating procedures. The choice of operating procedures, however, is logically independent desired rate of adjustment. That is. any the federal funds of the three most prominent types of operating of the procedures -- rate, nonborrowed reserves, and borrowed reserves procedures -- can potentially yield virtually the same rate of adjustment.5 Nevertheless, the adoption of a federal funds rate, or money market conditions, operating procedure is frequently interpreted as an abandonment of monetary targets. Similarly, the adoption of the nonborrowed reserves, or reserves aggregate, procedure in October 1979 is often viewed as being consistent with a greater desire to offset monetary shocks. As discussed in the next section, actual interest rate behavior can be used to infer both the type of operating procedure adopted by the Federal Reserve and the desired rate at which monetary shocks are offset. 4 II, Imolications for Interest Bii Market interest rate data are used to represent the actions of financial arket participants. It is further assumed that market participants fully understand Federal Reserve behavior. As a consequence, the Federal Reserves reaction function can be inferred under this rational expectations assumption. Moreover, changes in the estimated reaction function can be used to infer changes in monetary policy regimes. Three aspects of interest rate behavior are considered here. These are the volatility of interest rates, the response of interest rates to weekly MI announcements, and the response of interest rates to new information about inflation, economic activity, and exchange rates. Differences in the volatility of interest rates in different periods are used to infer changes in Federal Reserve operating procedures. Similarly, changes in the response of interest rates to Ml announcements also are used to infer changes in operating procedures as well as different degrees of emphasis on Ml targets. Finally, the response of interest rates to other economic information is used to infer the targets of monetary policy during a particular period. A. Volatility of Interest EI The volatility of the federal funds rate depends on disturbances affecting the market for reserves, Federal Reserve intervention in the reserves market through open market operations, and the market's perception of the type of operating procedure being used, If the market believes that the Federal Reserve will offset shocks affecting the reserves market through open market operations, and the shocks are in fact offset fairly quickly, the federal funds rate will be relatively stable over a short period, such as a week. If disturbances in either the demand for or supply of reserves are not expected to be offset, however, the federal funds rate will move quickly to clear the reserves market. Different operating procedures imply different behavior for the federal funds rate through these channels.6 Under the federal funds rate procedure, the Federal Reserve offsets most shocks affecting the reserves market to keep the federal funds rate relatively stable over a given period, such Even under this procedure, however, the federal funds rate would be expected ) as a week. to exhibit some volatility over time. In particular, to offset deviations in money growth from its target, the Federal Reserve may initiate discretionary changes in the rate, Nevertheless, in comparison to the other operating procedures discussed below, the federal funds rate should be relatively more stable under this procedure. Under the nonborrowed reserves procedure, moat disturbances affecting the reserves market, and therefore the federal funds rate, are not offset. Instead, the nonborrowed reserves path is maintained over a given period and the federal funds rate fluctuates in response to shocks either demand fur or supply to the of reserves, As a result, the federal funds rate would be expected to be more volatile under this procedure. The final procedure considered is the borrowed reserves procedure. Under this procedure, the Federal Reserve can be. characterized as attempting to' achieve a certain level of discdunt window borrowing over a given period. In this case, unanticipated changes in either required or excess reserves are accommodated by changing nonborrowed reserves, If the demand for required reserves is higher than expected, fur example, the federal funds rate rises initially and borrowing increases to equate supply and demand in the reserves market. To offset the increase in borrowing, nonborrowed reserves are increased until the federal funds rate falls to its previous level. In contrast, if a shock originating in the demand for borrowed reserves occurs in which borrowing is higher than expected at every level of the federal funds rate, this disturbance is at most partially offset and the federal funds rate falls. The decline in the federal funds rate serves to reduce the demand for borrowed reserves. So, this source of disturbances in the reserves market causes fluctuations in the federal funds rate. In comparison with the other procedures, the borrowed reserves procedure implies more short-run volatility in the federal funds rate than the federal funds rate procedure and less volatility than the nonborrowed reserves procedure.7 The volatility of other interest rates, such as the 3-month Treasury bill yield, also depends on the type of operating procedure employed by the Federal Reserve, although to a lesser extent. The 3month Treasury bill yield depends on both the current federal funds rate and the rate expected in future weeks. If the curreot week's federal funds rate fluctuates, then some of this volatility is reflected in the Treasury bill yield. Treasury bill yields also fluctuate if financial market participants change their assessments about the federal funds rate in future weeks. Monetary targets are important in examining this link. If new information about either money or the economy suggest8, for example, that the future level of the money stock will be higher than previously expected, the Treasury bill yield may rise if the market expects the Federal Reserve to offset this increase. In this instance, the market expects the Reserve to attempt to achieve a particular monetary target, and the magnitude of the Federal increase in interest rates reflects the desired speed of short-run adjustment.back to the target level. Alternatively, if the Federal Reserve places little or no weight on a particular monetary target, the market will expect future levels of the federal funds rate to be as previously predicted. So, for a given monetary disturbance, the greater the Federal Reserve's commitment to achieve a particular monetary target, the greater the coinciding fluctuation in longer term yields. B. Response to g Economic Announcements The response of interest rates to money and economic announcements provides futher evidence on the type of operating procedure adopted by the Federal Reserve, as well a.s the degree of emphasis placed on various targets and intermediate targets of policy. In particular, the response of the federal funds rate depends on the type of operating procedure. The response of longer-term yields depends somewhat on the operating procedure, but target or target of policy. The responses to it depends more importantly on the intermediate Ml announcements and other economic announcements are considered separately below. First, consider the response to the federal funds rate to weekly Ml announcements. Th response depends directly on the type of operating procedure employed by the Federal Reserve, In particular,the response depends on whether the corresponding shock to the market for reserves is offset. The reserves market is affected by unanticipated announced changes in Ml initially through the market's assessment of the demand for required reserves. Under the LRR system in effect before February 1984, required reserves depended on the level of the money stock two weeks previously, the statement week corresponding to the current week's money announcement data. Under the CRR system adopted in February 1984, required reserves depend on the current money stock, with a lag of two days.8 As a consequence, the money announcement data under current reserves periods. Unanticipated announced changes in CRR do not coincide with the Ml may still affect the demand for reserves, however, if the unanticipated changes have persistent effects on future levels of the money stock, That is, the current week's demand for reserves would he affected if a positive money announcement surprise causes market participants to raise their assessments of the current week's money stock. Under the federal funds rate procedure, the federal funds rate should not respond to money announcement surprises, in this case, market participantsexpect the Federal Reserve to accommodate the implied shock in the reserves market, In contrast, the federal funds rate should increase in response to a positive money announcement surprise under the nonborrowed reserves procedure. This rise is due to a higher assessment of the demand for reserves that is not expected to be accommodated through Federal Reserve open msrket operations. Similar to the federal funds rate procedure, the federal funds rate should not respond to money announcement surprises under the borrowed reserves procedure. This behavior follows because the Federal Reserve accommodates shocks to the demand for reserves under this operating procedure. The response of Treasury bill yields and other longer term yields to money announcement surprises depends partly on the response of the federal funds rate. Most of the response. however, depends on the extent to which the market expects the Federal Reserve to offset the shock in the future. In particular, the response is greater the more quickly the Federal Reserve acts to offset the money surprise. So, the response to the Treasury bill yield to money announcement surprises can be used tn determine whether the Federal Reserve is perceived to he attempting to achieve its Ml target.9 The response of Treasury bill yields to unanticipated announced changes in economic activity and inflation may operate through the same channels as money announcements if these other economic announcements provide information useful in predicting money demand. In particular, if either real economic activity or inflation is higher than expected, the market may raise current and future week's demand for money. its assessment of the interest rates, then, wnuld be expected to rise in response to this new information about the domestic economy 8 if the market does not expect the Federal Reserve to accommodate the increased demand. A 8imilar direct link between exchange rate movements and the demand for money is not evident. The primary effect of new information about economic activity, inflation, and foreign exchange rates on interest rates is likely due to the direct value of this information. That is, if policymakers are placing more weight on their targets, and less weight on their intermediate monetary targets, new information about the economy and exchange rates may cause immediate interest rate movements. If inflation is announced to be higher than expected, for example, policymakers may adopt more restrictive policies, causing interest rates to rise immediately. The effect could be the same for an unexpectedly large increase in economic activity, especially during the later stages of an economic expansion when concern about future inflation is growing. Moreover, if the Federal Reserve is focusing directly on the value of the dollar in currency markets, new information about exchange rates may lead to immediate interest rate movements. Depending on the Federal Reserves ernphass on this type of direct information about its ultimate targets, the response of interest rates may have varied over different periods. Ill. Empirical Results The various aspects of interest rate behavior discussed in the previous section are used below to test hypotheses about the behavior to the Federal Reserve in conducting monetary policy. Three separate periods are examined since 1980. The first actually begins in October 1979, when the Federal Reserve announced a new operating procedure along with an increased emphasis on the monetary aggregates. This new procedure corresponds to the nonborrowed reserves procedure discussed in the previous section. The second period begins in October 1982, when the Federal Reserve announced the abandoned the nonborrowed reserves procedure in favor of the borrowed reserves operating procedure. At that time, somewhat less emphasis also may have been placed on the monetary aggregates specifically Ml -- as intermediate targets. The — third period begins in February 1984, coinciding with the implementation of CRR. Because of the uncertainties regarding the effects of CRR among other factors, the Federal Reserve may have dc-emphasized the role of the monetary aggregates further in 9 this period.10 Another period, beginning in October 1977 and ending in October 1979, also is examined to compare the monetary policy regimes in the 1980s with monetary policy in the late 1970s. Prior to October 1979, the federal funds rate procedure was in effect. A. Volatility of Interest The volatility of the federal funds rate and the 3-month Treasury hill yield is examined over four periods in Table 1. As is apparent in the table, the weekly volatility of the federal funds rate TABLE I is consistent with the changes in operating procedures hypothesized previously. in particular. in cbntparisun tu the pre-October 1979 period, the standard deviation of weekly percentage changes in the federal funds rate was about four times as large in the October 1979 - October 1982 period. This behavior is consistent with a switch to the nonhorrowed reserves procedure from the federal funds rate procedure. The volatility then declined to about half that of the October 1979 - October 1982 period in both of the periods after October 1982. Despite this decline, the volatility since October 1982 is significantly larger than that experienced prior to October 1979, suggesting that the Federal Reserve did not return to the federal funds rate procedure.1' adoption Instead, the evidence is consistent with the of the borrowed reserves procedure. Similar to the volatility of the federal funds rate, the volatility of the 3-month Treasury bill yield also increased significantly following October 1979. In contrast to the federal funds rate, however, volatility after October 1982 returned to pre-October 1979 levels. While some portion of the volatility of the Treasury bill yield should reflect the volatility of the federal funds rate, a larger part can be attributed to changes in the market's expectation about future monetary policy, As a result, the increased volatility of the Treasury bill yield in the October 1979 - October 1982period is consistent with a greater commitment of the Federal Reserve to offset a given shock affecting the money stock, while the decline in volatility after October 1982 suggests the opposite. 10 Other evidence, however, allows more direct inference about this possible shift. B. Response to MQ! Economic Announcements To provide further evidence on changes in monetary policy operating procedures, hypotheses discussed in the previous section about the response of the federal funds rate to money announcement surprises are first tested. Then, to determine the market's assessment of the Federal Reserve's targets and intermediate targets, the response of the 3—month Treasury bill yields to money and other economic announcements is examined empirically. The usual efficient markets model is used rate and the 3-month Treasury bill yield. to estimate the responses of both the federal funds This model relates daily changes in interest rates to unanticipated announced changes in money and other economic variables. With the exception of the exchange rate variable, unanticipated changes are measured as the difference between announced and expected values. The expected values are taken from a survey conducted by Money Market Services. Inc. The survey data are further adjusted, however, to take into account information from the time of the survey to the time of the announcement.'2 views new information about a particular If the Federal Reserve -- and therefore the market -- variable as being relevant for policy, the estimated response of interest rates should be significant. The estimated response of the federal funds rate to the unanticipated component of weekly Ml announcements is reported in Table 2. The only statistically significant response occurs in the January TABLE 2 198Q - October 1982 period, consistent with the nonborrowed reserves procedure.13 In the two post- October 1982 periods, the response is insignificantly different from zero, consistent with either the federal funds rate or borrowed reserves procedures. Although estimates are not reported here, previous studies find that the response also is insignificantly different from zero prior to October 1979 [e.g., Roley and Walsh (1985)]. Combined with the evidence from Table 1, the results support the hypothesis that the Federal Reserve operated under a federal funds rate procedure prior to October 1979, and then implemented the nonborrowed reserves procedure 11 during the October 1979 - October 1982 period. After October 1982, the combined evidence from Tablss 1 and 2 suggests that the Federal Reserve adopted the borrowed reaerves procedure. To determine the relative importance of various targets and intermediate targeta of monetary policy, the response of the 3-month Treasury bill yield to a set of economic variables is estimated in Table 3. In addition to weekly Ml announcements, the unanticipated components of monthly annnuncements of the producer price index, consumer price index, industrial production. and the unemployment rate are considered. Daily changes in the yen/dollar exchange rate also are included to estimate the significance of foreign exchange rate factors. Yen/dollar rate changes are measured as the difference in ciosing and opening quotes in the Tokyo market each day. Thus, such movements in the Tokyo market cao be regacded as news to traders in New York, since these movements are recorded when the New York idarket is ciosed. TABLE 3 The results in Table three periods since 3 indicate that the role of Ml targets successively diminished over the 1980. and the differences across periods are significant at the 5 percent level. Moreover, since February 1984, the response of the 3-month Tressury bill yield to weekly Ml announcements is insignificant!y different from zero'4 While the role of Mi has decreased, the reeulte suggest that the effect of direct information about the domestic economy has increased over time. As indicated by the reported F-statistics, the hypothesis that new information about inflation, unemployment, and output does not affect interest rates can be rejected February 1984 - September at the 5 percent level in the 1986 period. The most significant variable during this latter period is industrial production, suggesting that the Federal Reserve was looking directly at the performance of the economy in conducting policy. Finally, yen/dollar exchange rate movements are not estimated to have had significant effects overall, though intermittent effects cannot be ruled 12 out.' IV. Conclusions This paper exploited the financial market's understanding of Federal Reserve behavior in considering changes in monetary policy. In particular, both interest rate volatility and the financial market's perception of the Federal Reserve's reaction function were osed to examine several monetary policy regimes over the last decade. One regime began in October 1979, when the Federal Reserve a nonhorrow"d adopted October 1979, the reserves operating procedure and placed greater emphasis on Ml. fedrai funds rate operating Prior to procedure was to effect. In October 1982, monetary policy shi1ted to a borrowed reserves operating procedure and a diminished role for Mi. Finally, in February 19k4, stil less weight was placed on Ml as an intermediate target of policy, with more weight placed directly or. the performance of the domestic economy These d,fferent periodc correspond to Federal Reser.e start-ri.rnts about changes in policy, and, as a whole, the evidence suggests tha actual changes wet" n.ade. To that extent. tne evidence suggests that piliry was credible. To anayce the motives surrounding the Federal Reserre's formulation and inplcnientation of niooetary 1eAtry, it seems particularly worthwhile to study these specific episodes on a case study basis. Such a sun> nay reseal the relative importance placed on political and economic factors influencing Federal lleser'.e behasior. In brief, it appears that both factors are important. 'The change in policy in October 197 for exa'nple. was drisen by the desire to ncrcase interest rates to defend the dollar in foreign eamhange narkets and to reduce inflation in the U.S. A slowdown in the growth of the monetary aggregates also was desired to reduce infiatioo in the future. Under the federal funds rate procedure in effect prior to October 1979. the Federal Reserve would be blamed for interest rates. a sharp rise in To deflect such criticism, policymakers adopted a procedure in which "the market determines rates," the nooborrowed reserves procedure. At the same time, there was some hope that this procedure would allow closer control of Ml and hence reduce the prospects of continued high inflation. Similar economic and political choices were made in 1982. In particular, by October 1982, the behavior of both Ml and the economy was seemingly becoming more unpredictable, and the Federal Reserve's independence was being threatened by the Congress. Policymakers decided to weakeo the 13 link between Ml and interest rates, thereby reducing interest rate volatility Nevertheless the advantage of some interest rate volatility due to market factors was the lack of blame attached to the Federal Reserve for a particular level of interest rates. As a consequence, the borrowed reserves procedure was adopted, in which the federal funds rate fluctuated in response to errors in the borrowings function. targeting In this sense, the Federal Reserve could continue to claim that it was not a particular level of interest rates. A macroeconomicjustification for the borrowed reserves procedure is not, however, readily apparent, as the added volatility of interest rates created by random shifts in discount window borrowing creates no clear benefit.16 The third regime in the I980s appears to have based on the uncertainties f been due primarily to an economic decisioo the economic environment, Jo particular, the implementation of CRR in February 1984 had the potential of affecting the behavior of an already unpredictable money stock as well as altering the previous rslatiooships between reserves aod money. This change was a delayed product of the nonborrowed reserves regime, as closer monetary control was more likely under CRR than LRR. Because of the added uncertainties surrounding CRR and the continuing erratic behavior of Ml velocity, policymakers dc-emphasized their monetary targets further, and the borrowed reserves procedure was maintained. In sum..t e behavior of the Federal Reserve in conducting monetary policy is best described by considering the Federal Reserve's operating procedures, targets, and desire to achieve its targets. Changes in these factors led to at least three different policy regimes in the 1980s.' Jo these regimes the importance to the Federal Reserve of tight monetary control increased initially and subsequently became negligible. The lasting inheritance of the regime shift that began in October 1979 is the Federal Reserve's substantiallygreater willingness to tolerate interest-rate volatility. Regardless of the specific details of the operating procedures adopted, there apparently has been no reversion to the federal funds rate procedure employed in the late 1970s. 14 FOOTNOTES * We are grateful to Thomas Mayer for helpful comments, to Steven R. Thorley for research assistance, and to the National Science Foundation (Grant No, SES-8408603) for research support. 1. The terminology used here for targets, intermediate targets, and instruments of policy follows B. Friedman (1975). 2. Mayer (l97) and Goodfriend (1986) discuss issues relating to the disclosure of the FOMC's policy directive in detail. 3. One example among many possible illustrations, wo.ild be the avoidaoce of confusion between use of an interest rate instrument to achieve a particular policy target and interest rate targeting where the level of an interest rate itselfis the policy objective. 4. For further discussion on tins issue, see LeRoy (1979. Hetcel (1982), and Roley (1987). 5. The nonborrowed reserves procedure is potentially slightly faster because of federal funds rate moserncnts in the current week. This property is discussed further in the next section. Also see Roley 1987). 6. For a more detailed discussion of these operating procedures, see Roley (1987). 7. A variant of the federal funds rate procedure which apparently was introduced following the sharp decline of the stock market on October 19, 1987 is intermediate between the federal funds rate procedure described above and the borrowed reserves procedure. In contrast to the earlier funds rate procedure, in which the Federal Reserve entered the reserves market almost daily to add or drain reserves in order to keep the funds rate close to target, open market operations under the variant procedure are relatively sparse and generally limited to adding or draining reserves, but not both, during a reserves maintenance period. variant resembles the borrowed reserves procedure. The main difference In this regard, the is that there is no explicit target for borrowed reserves, so that shocks to banks' demand for borrowed reserves do not (in principle) affect the funds rate. Some market participantshave described the variant 15 procedure as a "fuzzy" funds rate target, since the allowable deviation the target is larger of the actual rate from than under the earlier funds rate procedure. The Federal Reserve has announced its intention to revert to the borrowed reserves procedure at some point in the future. 8. Also, reserve computation and maintenance periods are two weeks in length, while they lasted one week under LRR. The response under 2-week CRR is examined in detail by Roley (1987). 9. Considerable evidence suggests that the response of Treasury bill yields to Ml announcements is a response in the real rate, not expected inflation. One type of evidence is from the response of foreign exchange rates to Ml announcement surprises. In particular, the dollar appreciates in response to positive money announcement surprises. See, for example, Cornell (1982), Engle and Frankel (1984), and Flardouvelis (1984). 10. - The introduction of CRR introduced describing at least two uncertainties. First, the previous relationship the demand for reserves changed, leading to uncertainties about the effects of various shocks on the federal funds rate. Second, the stochastic behavior of the money stock itselfwas potentially affected, as it was no longer entirely demand determined in a given week. Ii. The test of the hypothesis that the two post-October 1982 periods have the same volatility as the pre-October 1979 period yields F-statistics of 3.88 (64, 100) and 5.29 (154, 100), which are significant at the 5 percent level, 12. The change in the Treasury bill yield over the previous five business days is used as a proxy for this information. See Roley (1983). 13. The results are qualitatively the same when the estimation period starts in October 1979 instead of January 1980. See Roley and Walsh (1985). To conform with the results in Table 3, the January 1980 starting period was used. This date was chosen because of the availability of data for the exchange rate variable used in Table 3. 14. In a specification analogous to that used in Table 2, the response of the Treasury bill yield to money announcement surprises also was examined over additional subperiode. In particular, starting with September 29, 1977, the response was estimated for overlapping 26 week periods, 16 beginning every 13 weeks. The first two estimation periods, for example, were September 29, 1977 - March 23, 1978 and December 29, 1977 - June 22, 1978, respectively. The results from these regressions generally support the beginning and ending dates of the subsarnplea used in the tables. In particular, all responses after the June 1982 - December 1982 period were estimated to be smaller than those of this period. Moreover, the responses in both the June 1983 - December 1983 and September 1983 at March 1984 periods were statistically significant the 5 percent level, but the esttrnated responses starting with the Decerriber 1983 1984 period were not significant. This result is consiscent with - June a change in thc response around the beginring of 1984, 15. Positive ar,d negative movements also were considered separately, but the statistical significance of the revolts did 16. nt clang" The borrowed reserves procedure was implemented under LRR. Under CRR, hos,eer, this procedure -nay be capable of offsetting shocks from money and factors supplying reserves. See, for example, Roth and Seibert '1983). 17. And, as noted in footnote 7, a fourth regime may have begun following the October 19, 1987 stock-market crash. 17 REFERENCES Cornell, Bradford, "Money Supply Announcements, Interest Rates, and Foreign Exchange," International and Finance, 1 Mth1 91 (August 1982), 201-208. Eogie, Charles, and Jeffrey Frankel. "Why Interest Rates React to Money Announcements: Explanation from the Foreign Exchange Market" of ar An Economics, 13 (January 1984). 13-39. Friedman; Benjathin M., "Targets, Instruments and Indicators of Mooetary Policy," 1913m91 of Monetary Economics; 1 (october 1975), 443-473. Secrecy and Central Banking," Journal 91 Monetary Goodfriend, Marvin, "Monetary Mystique: Economics,17 (January 1986), 63-82. Hardouvelis, Olkas, "Market Perceptions of Federal Reserve Policy and the Weekly Monetary Announcements," Journal of Monetary Economics. 14 (September 1984), 225-240. jt Hetrel, Robert F., "The October 1979 Regime of Monetary Control and the Behavior of the Money Supply in 1980," Journal of Money, and fijün . 14 (May 1982), 234-251. LeRoy, Stephen F., "Monetary Control Under Lagged Reserve Accounting," 1l Q3jrn Economic (October 1979), 460-470. Mayer, Thomas, "Disclosing Monetary Policy," Saloman Brothers Center for the Study of Financial Institutions, New York University, Monograpri 1987-1. IS Roth, Howard L. and Diane Seibert, 'The Effects of Alternative Discount Rate Mechanisms on Monetary Control," Federal Reserve Bank of Kansas City, Economic 68 (March 1983), 16-29. Roley V Vance. The Response of Short-Term Announcements," Journal of Money. Ro1e V. Vance, "The Effects Procedures,' Role Jpa of V. Vance, and Carl E. of (Supplement to Weekly Money Supply and Banking, 15 (August 1983), 344-354. of Money Announcements under Alternative Monetary Control Money, and Bankme, Wah, "Monetary Interet Rate to Interest Hates Money 19 (August 1987), 292-307. Poitcy Regimes, Expected Inflation, and the Response Announcements," Quarterly 19135), 1011-1039. 19 Journal of Economics. 100 TABLE 2 Response of the Federsl Funds Rate to Ml Announcements Coeffioient Estimates Estimation UM1 41 0.0994* —0.0121 Period Constant jan, 1980— Oct. 1982 (0.0515) Oct. 1982Feb. 1984 0.0413 0.0137 0.0016 (0.0272) (0.0128) (0.0122) _0.O872* 0.0103 -0.0089 (0.0191) (0.0103) (0.0055) 0.0277 Feb. 1984Feb. 1987 Notes: (0.0228) Summary Statistics — R2 SE OW 0.11 0.59 2.70 -0.01 0.20 1.76 F(m,c) (0.0291) 10.71* (1,203) 0.01 0.22 1.97 0.04 (1,219) 5 The preoise estimstion period dstes are: Jsnusry 1, 1980 — October 1982; Ootober 6, 1982 — Februsry 1, 1984; and February 2, 1984 — February 28, 1987. Obsenstions in the last week of each year are deleted to avoid the influence of excessive year end interest rate movements. Numbers in parentheses sre stsndsrd errors of estimated coefficients. Equations sre estimated in the form: LRFF 0 b. where b term. , , t =b0 +b1 and b 2 UM1 t +bEMl +e t 2 t are estimated coefficients and e 8 ía a random error * Srgnificant at the 5 percent level. = change in the fedaral funds rate from the day of the money announcement 4RFFt to the next business dsy. UM1 8l, money announcement surprise, defined as Ml — where Mi is the anncunccd change in the narrowly defined money stock, in billions of dollars. SE = expected announced chsnge in the narrowly defined money stock, based cc the survey measure provided by Money Market Services, Inc. = multiple correlation coefficient corrected for degrees of freedom. standard error. OW = Ourbin—Wateon statistic. F(m,m) F—statistic with (m,:) degrees of freedom for the hypothesis that the response coefficient, b1, is the same ss that estimated in the previous period. In this tet, the estimated equations sre weighted by their standard errors. 20 TABLE 1 Volatility of Interest Rates Standard Deviations of Weekly Percentage Changes Oct. 1982— Feb. 1984— Oct. 1979— Oct. 1932 Feb. 1984 Feb. 1937 Interest Rate Oct. 1977— Oct. 1979 Federal Fonda rate 0.019 F(m,n) 0.077 * 16.89 (150, p—value 3—month euty 4.36 (150, 64) 0.1 a l02 0.043 1.37 * (154, 64) 0.05 0.021 0.05: 4,62 0.02: 99 '150, 64) (150, u.2 a L0 p—value 5.08 0. s 10-15 1.53 '54, 154) a The intereat rate quote on the day following each week's money announcement The number in the f'rst row and is used to form weekly percentage changes. first coiumn, for example, denotes s standard dewiation of weekly percent— age cionges of 1.9 percent. The last weekly observat:on of each year is deleted to avoio the Influence of excessive year end interest rate movenenta. Significant at the F(m,n) * 0.027 F(m,n) Notes: 99) 0.00 0.037 5 percent level. F—statistic with degrees of freedom for the hypothesis that the varithat in the previous period. (m,n) ance is the same as 21 Ml YEN j2 SE 0.02 0.08 1.89 21.95* (1,992) 0.0019 —0.0069 (1,1026) —— (4,662) 4,85* 1.34 (4,330) (4,694) 0.48 F2(m,n) Tests _____ F1(m,n) (0.0029) (0.0443) (0.1066) (0.0431) (0.0689) (0.0043) (0.0051) 0.1820* —0.0800 0.0305 —0.1964 1.70 DW Statistics percentage of labor force unemployment. percentage change in the industrial production index. percentage change in the consumer price index. F--statistic with (m,n) degrees F2(m,n) on day t. of freedom for the hypothesis that b1 h2 = b4 0. with (m,n) degrees of freedom for the hypothesis that the response coefficient, b5, is the same as that estimated in the previous period. The relevant estimated equations are weighted by their standard errors, F—statistic = change in the yen/dollar exchange rate from open to close in the Tokyo market = change in 3—month Treasury bill yield from 3:30 p.m. to 33O p.m. on auccessive business days = percentage change in the producer price index, F1(m,n) = YEN UNEM IP CPI RTBt PPI where the U's indicate that only the unanticipated components of the data are included. Unanticipated values of the first five variables are calculated using survey data provided by Money Market Services, Inc. ARTB = b0 + b1 UPPI + b2 UCPIL + h3 un + b4 UUNEMt 1- b5 UMl + b6 YENt + Notes: See the notes in Table 2. The last period ends on September 29, 1986, due to the availability of yen/dollar exchange rate data. Equations are estimated in the form: Sept.1986 lINEN SuTy 0.11 0.09 1.55 16.59* IP 0.0033 —0.0921 —0.2048 0.0558 —0.1758 0,0346* 0.0100 (0.0047) (0.0763) (0.1651) (0.0751) (0.1358) (0.0055) (0.0067) 0.0290 —0.2048 Feb. 1984— 0,0058 —00075 —0.0056 Feb. 1984 Oct. 1982— CPI —_Unantic4dChaifn 0.07 0.28 PPI 0.0842* 0.0029 (0.0107) (0.1620) (0.1929) (0.1084) (0.2438) (0.0109) (0.0112) Constant 1980— —0.0079 Oct. 1982 Jan. Period Estimation Response of the 3—month Treasury Bill Yield to New Economic Information TABLE 3